Implementation of Hybrid ARIMA-ANN and ARIMA- Local Prediction Techniques in the Traffic Management System. Case Study: Airport of a Metropolis

Nowadays, time series prediction is one of the fundamental research purposes, owing to the importance of prediction in various real-world applications. Transportation management is one of the main issues of each municipality that needs prediction. Accurate and reliable forecasting is one of the fundamental goals of an intelligent transportation system. In the literature, it is defined several variables of traffic management such as speed, time, and flow. In the proposed paper, a case of an Airline is considered. The associated managers are planning to propose a new service for passengers. There is a need to predict the number of passengers on arriving flights to a metropolis in Iran. This variable is inherently similar to the flow in traffic management. In the literature on traffic management, most of the studies implemented a linear or nonlinear modeling method to predict the future and ignore the advantage of hybrid methods. Several hybrid ARIMA-ANN methods have been proposed to specify the underlying relationships among the data. This paper utilizes a hybrid ARIMA-ANN model which decomposes the data into lowvolatile, and high-volatile components to predict accurately. Also, the current paper develops a new hybrid method, ARIMA-Local method, to specify the efficiency of other provided nonlinear methods in a hybrid structure. The obtained results for the discussed case are reported. This study signifies the accuracy of ARIMA-ANN model in predicting, while also the ARIMA-Local method is efficient in forecasting in comparison to the individual models of ANN and Exponential smoothing.